   Mn_Fit version 4.01/21 is hereby released. This has small bug fixes
with respect to 4.01/16: `Set colour' with a colour number was not
right, circular symbols had the wrong colour for the rim sometimes,
`Show tick' and 'Show label' had small bugs and `Display' or
`Fit_info' of 1000 bin histograms caused a divide by 0.
HBOOK histograms are now all fetched with hunpak and hunpke. This
should be faster than using hi and hie.
In addition commands like set x label '(GeV/c^2!)' do not work
interactively with readline verisons, because ! is used for history.
You must give the command set x label '(GeV/c^2\!)'. This meant a fix
to typscn also to substitute `\!' with `!'.

    Version 4.01/16 has command recall and history for all Unix
machines using GNU readline routines. It is adapted to the new kuip
condition handling routines.  This means that the main routines are
written in C. In addition I have included the necessary kuip include
files in the Mn_Fit cmz file.  This means that the version of the CERN
library that is being used must be given.

    This version also runs on Alpha machines. I assume that the symbol
ARCHITECTURE="ALPHA" exists on Alpha machines.

    There are now install scripts for Vax, Alpha, Apollo, HP,
Decstations and SGI. The scripts have been used on Vax, Alpha, Apollo,
HP and Decstations. Scripts for other Unix machines can be generated
trivially. I just need to know if the machine has bsd or sys5 flavour
of unix. These scripts are also included in the standard directories.

    Note that this version of Mn_Fit is only compatible with version
93d of the CERN libraries. This is because there are some changes in
the syntax of a few CMZ commands that I use.

    For Unix machines I have a compressed tar file of the Mn_Fit code
and the hp executable. This can be obtained from the directory
/user/brock/mn_fit/tar. The following files are there:

  mn_fit.tar.gz               Mn_Fit source and install scripts
  mn_fit_x.exe.hpux.tar.gz    Mn_Fit hpux X windows executable + other files
  mn_fit.ps.tar.gz            Mn_Fit manual

  To unpack one of these files do the following:

  cd $MN_FIT                  (or cd the Mn_Fit top-level directory)
  gunzip filename.tar.gz      (or gzip -d ...)
  tar xvf filename.tar

  If you get mn_fit.tar.gz you will have the Mn_Fit car files, the
  install scripts and GNU readline compressed tar file. You can then
  run the appropriate install script.

    The Mn_Fit source code can be obtained from the following
directories:

Vax:          vxl3j0::l3daq$dat:[gra.mn_fit.pro]
              vxcern::disk$scratch:[l3.week.brock.mn_fit.pro]
              lns62::ud:[mn_fit.pro]
              cmphys.phys.cmu.edu       col$root:[brock.mn_fit.pro]
Apollo:       CERN                      /user/brock/mn_fit/pro
Hp:           CERN                      /user/brock/mn_fit/pro
IBM:          cernvm                    brock.499
Decstations:  fermi.phys.cmu.edu        /users/brock/mn_fit/pro
              lns717.lns.cornell.edu    /cdat/tem/mnfit/pro

Note that the files on vxcern may disappear after a while as I do not
logon there very regularly.

    The following files should be copied from these directories:

mn_fit.car
mn_util.car
typscn.car
minuit.car
install.vms                  -> install.com       (VMS)
install.hpux                                      (HP)
install.decs                                      (Decs)
install.sgi                                       (SGI - not tested)
install.apollo                                    (Apollo)
mn_fit.doc
mn_readline-1.1.tar.gz                            (Unix machines)
mn_readline-1_1.tar_gz                            (from VMS for Unix)
readline.tar_gz                                   (from CERNVM for Unix)

These directories will saty there for as long as I regularly login to
vxcern!

    If you get the files by ftp then you should do a binary get on
mn_readline-1.1.tar.gz.

    For Unix machines I also have a compressed tar file that I can
mail to you if you cannot access any of the nodes given above. If I
mail you this file you need to do the following:

From mail:

Save the mail to a file in the Mn_Fit top-level directory (or /tmp).
e.g. in elm: s
             $MN_FIT/mn_fit.tar.gz.mail or /tmp/mn_fit.tar.gz.mail

Outside mail:

  cd $MN_FIT                  (or cd the Mn_Fit top-level directory)
  uudecode mn_fit.tar.gz.mail (or /tmp/mn_fit.tar.gz.mail)

Then follow the procedure given above to unzip and detar the file.

    The install procedure should be fairly self-explanatory. On Unix
machines I assume that you have cmz. In theory it is possible to get
all the Fortran on VMS, but I have not tried that option.

    More documentation can be found in mn_fit.doc.


Vax and Alpha machines:
_______________________


The standard way to define the cmz command on VMS is:

cmz == "@cern:[pro.exe]cmz"

    This does not work with cmz/batch. You must redefine cmz to be
something like:

cmz == "$cern:[cmz.pro.exe]cmz.exe"

Decstations:
____________


    For the moment I have had some difficulty compiling the c version
of the Mn_Fit main program on Decstations. I have changed the c
compiler command to c89 and this works fine at Cornell.  If you do not
have c89, then you must change the command to cc in
//mn_util/$kumacs/mn_macro and also change mn_fit.lnk to include
mn_main.o instead of mn_main_cc.o there. Please let me know if you
still have problems.

Apollo:
_______


    Note that the command recall version does not work properly in
Apollo pads. I therefore check the TERM environment variable on the
Apollo and if it is "Apollo" then I run the non-readline version. This
means doubling the number of executables on Apollos.

                                 Regards
                                   Ian

